Thesis (PhD)--Stellenbosch University, 2005. / ENGLISH ABSTRACT: To maximise the throughput of a production system the capacity constrained resource needs to be
protected from variation and uncertainty. In the Theory of Constraints philosophy such protection is
provided by means of time buffers and protective capacity. Time buffers are protective time that is
allowed in the production schedule to buffer against disruptions, whereas protective capacity is
defined as a given amount of extra capacity at non-constraints above the system constraint's capacity.
In this research an analytical procedure was developed to more accurately determine the required time
buffer lengths. This procedure uses an open queuing network modelling approach where workstations
are modelled as GIIG/m queues. A simulation experiment was performed to evaluate the time buffer
estimation procedure on the operations of an actual fifteen station flow shop. The results from the
study suggest that the analytical procedure is sufficiently accurate to provide an initial quick estimate
of the needed time buffer lengths at the design stage of the line.
This dissertation also investigated the effect of protective capacity levels at a secondary constraint
resource as well as at the other non-constraint resources on the mean flow time, the bottleneck
probability of the primary constraint resource, as well as the output of flow production systems using
simulation models and ANOV A. Two different types of flow production systems were investigated:
(1) a flow shop with a fixed number of stations and unlimited queue or buffer space between stations,
and (2) an assembly line where a total work content is distributed among stations in a certain fashion
and the number of stations are not fixed.
The experimental studies show that flow shop performance in the form of flow time and line output is
not that much influenced by low protective capacity levels at the secondary constraint resource. Low
protective capacity levels at a single station however can significantly reduce the bottleneck
probability for the primary constraint resource when it is located before and relatively close or near to
the primary constraint in the process flow, or after but relatively far from the primary constraint. An
after-far secondary constraint location also causes slightly longer job flow times, and should therefore
be avoided if possible. The research further shows that quite high protective capacity levels at the nonconstraint
resources are needed to ensure a more stable and therefore manageable primary constraint.
However low average levels of protective capacity at non-constraint resources are sufficient to ensure
that the maximum designed output level as determined by the utilisation of the primary constraint
resource is obtained. The results for the assembly line experiment showed that an unbalanced line configuration where less
work is assigned to the non-constraint stations than to the primary constraint station (but nonconstraint
stations have an equal work content) can lead to significant reductions in the mean flow
time while maintaining the same line output, without resulting in too many additional stations. Low
protective capacity levels in the range of 2% to 5% are sufficient to cause substantial improvements in
flow time without resulting in too many additional stations in the line. / AFRIKAANSE OPSOMMING: Om die finale uitset van 'n produksiestelsel te maksimeer is dit noodsaaklik dat die bottelnek beskerm
word teen fluktuasies en onderbrekings. In die "Theory of Constraints" filosofie word van twee soorte
beskermingsmeganismes gebruik gemaak: tydbuffers en beskermende kapasiteit. Tydbuffers is
beskermende tyd wat in die produksieskedule gevoeg word om sodoende die bottelnek teen
onderbrekings te beskerm, terwyl beskermende kapasiteit ekstra produksiekapasiteit relatief tot die
bottelnek se kapasiteit is wat by nie-bottelnekke gevoeg word.
In hierdie navorsing IS 'n analitiese prosedure ontwikkel om meer akkurate berekenings van
tydbuffergroottes te verkry in produksiestelsels wat volgens 'n "Theory of Constraints" filosofie
bestuur word. Die prosedure maak gebruik van oop toustaan netwerk modellering waar werkstasies
gemodelleer word as GIIG/m toue. Die analitiese prosedure is ge-evalueer met 'n simulasie
eksperiment op 'n werklike vyftien stasie vloeiwinkel. Die resultate dui aan dat die analitiese
prosedure akkuraat genoeg is om vinnig aanvanklike beramings vir die benodigde tydbuffergroottes
tydens die ontwerpsfase van die produksiestelsel te verskaf.
Verder is ook ondersoek ingestel na die effek van beskermende kapasiteitsvlakke by die sekondêre
bottelnek asook die ander nie-bottelnekke op die gemiddelde deurvloeityd, die totale uitset, asook die
bottelnek waarskynlikheid vir die primêre bottelnek in vloei produksiestelsels deur gebruik te maak
van simulasie modelle en ANOVA. Twee verskillende tipes vloei produksiestelsels is ondersoek: (1) 'n
vloeiwinkel met 'n vaste aantal stasies en 'n onbeperkte buffer spasie tussen stasies, en (2) 'n
monteerlyn waar 'n totale werksinhoud op 'n bepaalde wyse onder stasies verdeel moet word en die
aantal stasies nie vas is nie.
Die eksperimentele studies dui aan dat die deurvloeitye en totale uitset van 'n vloeiwinkel me
noemenswaardig beïnvloed word deur lae beskermende kapasiteitsvlakke by die sekondêre bottelnek
nie. Hierdie maatstawwe word meer beïnvloed word deur die gemiddelde beskermende
kapasiteitsvlakke by al die nie-bottelnekke. Lae beskermende kapasiteit by 'n enkele werkstasie kan
egter die bottelnek waarskynlikheid vir die primêre bottelnek aansienlik verlaag indien dit voor en
relatief na aan die primêre bottelnek in die prosesvloei geleë is, of na, maar relatief ver, vanaf die
primêre bottelnek. 'n Sekondêre bottelnek ligging na maar relatief ver vanaf die primêre bottelnek in
die prosesvloei veroorsaak ook langer deurvloeitye, en moet dus vermy word. Verder dui die
navorsing aan dat redelike hoë gemiddelde beskermende kapasiteit by nie-bottelnekke benodig word
om 'n meer stabiele primêre bottelnek te verseker. In vloeiwinkels met lae vlakke van variasie en
onderbrekings is egter lae gemiddelde vlakke van beskermende kapasiteit by nie-bottelnekke voldoende om te verseker dat die maksimum ontwerpte uitset soos bepaal deur die benutting van die
primêre bottelnek behaal word.
Die resultate vir die monteerlyn eksperiment dui aan dat 'n ongebalanseerde lynkonfigurasie waar
minder werk aan die nie-bottelnek stasies as aan die primêre bottelnek stasie toegeken word (maar niebottelnek
stasies het 'n gelyke werksinhoud), aansienlike verlagings in deurvloeityd teweeg kan bring
terwyl dieselfde lyn uitset behou word. Dit is moontlik sonder te veel addisionele stasies in die lyn.
Die eksperimentele resultate dui aan dat lae beskermende kapasiteitsvlakke van tussen 2% tot 5%
voldoende is om beduidende verlagings in deurvloeityd teweeg te bring sonder te veel addisionele
stasies.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:sun/oai:scholar.sun.ac.za:10019.1/49782 |
| Date | 04 1900 |
| Creators | Louw, Louis |
| Contributors | Page, D. C., Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. |
| Publisher | Stellenbosch : Stellenbosch University |
| Source Sets | South African National ETD Portal |
| Language | en_ZA |
| Detected Language | English |
| Type | Thesis |
| Format | 141 p. |
| Rights | Stellenbosch University |
Page generated in 0.003 seconds